HOMOGENEITY OF GRANITE FABRICS AT THE METRE AND DEKAMETRE SCALES

Transcription

1 HOMOGENEITY OF GRANITE FABRICS AT THE METRE AND DEKAMETRE SCALES Philippe OLIVIER, Michel de SAINT BLANQUAT, Gérard GLEIZES and Denis LEBLANC Equipe de Pétrophysique et Tectonique, UMR 553 CNRS Université Paul-Sabatier, 38 rue des Trente-Six-Ponts, Toulouse, France ABSTRACT Magnetic fabrics of biotite-bearing granites were systematically determined, at the metre and dekametre scales, in three plutons previously studied for their overall magnetic structures, in order to characterize the spatial homogeneity and variability of the fabrics. These granites, with typically magmatic microstructures, have different mean magnetic anisotropies (P%): Sidobre (southwest Massif Central of France; P%=2.3), Bassiès and Trois-Seigneurs (French Pyrenees; respectively P%= 3.3 and P%=5.). In each site, two grids of 50 oriented specimens each, respectively one and ten metres apart from each other, have been studied in detail. The directional data, especially the lineations, strongly cluster around their means and have similar orientations on both scales. In map view, the fluctuations of these data are generally gradual and tend to form sigmoids but no clearly defined pattern, such as a C/S system was observed. The magnetic anisotropy and the bulk susceptibility are homogeneous as a whole, and display spatial organizations with no simple relationships with the structures. These results confirm, however, the validity of the homogeneous structural patterns obtained from entire plutons. INTRODUCTION Magmatic structure of granitic plutons is the topic of many recent studies, based either on direct measurements (e.g., Guineberteau et al., 187; Pesquera and Pons, 10; Gasquet, 12; John and Blundy, 13), or on the Anisotropy of Magnetic Susceptibility (AMS) measurements (see Bouchez, this volume). Many studies based on AMS point out the homogeneity of magmatic structures over wide areas and therefore permit better constraining of the modes of emplacement proposed for the studied plutons. In all these studies, the density of sampling is rather low, usually one site per km 2, which raises several questions, such as: i) does the homogeneity observed at the kilometre scale also exist at the lower scales and with which characteristics? ii) how does the fabric varies from one site to the next? iii) how representative is a site characterized by a few samples? To answer these questions, the magnetic fabrics of three different granites, chosen

2 114 Ph. OLIVIER ET AL. for their different fabric intensities and grain sizes, have been studied on the scales of metre to several tens of metres, particularly focusing on the characterization of the spatial homogeneity and variability of the fabrics. METHODOLOGY MAGNETIC FABRIC MEASUREMENTS The granites studied here belong to the group of the ilmenite-bearing granitoids (Ishihara, 177; Bouchez, this volume), for which the iron-bearing silicates, in this case biotite, induce a dominantly paramagnetic behaviour, as confirmed by the low magnitudes of magnetic susceptibility, ranging from 5 to 50 x 10-5 S.I. In these rocks a variation of one 10-5 S.I. unit corresponds to a variation of about 0.15% in iron content (Gleizes et al., 13). Magnetic susceptibility measurements performed at low field (3.8 x 10-4 T, 20 Hz, Kappabridge KLY-2 apparatus) on 11 cm 3 cylindrical rock samples, allow us to define the intensity and direction of the three principal axes k1 k2 k3 of the ellipsoid of susceptibility anisotropy. Tensor means of the axes corresponding to the n samples measured at a site are K1 K2 K3. Km, the mean susceptibility, is the arithmetic mean of Ki (i=1, 3). K1, the long axis of the anisotropy ellipsoid, is the magnetic lineation, K3, the short axis, is the normal to the magnetic foliation. Uncertainty on susceptibility measurement may be estimated at a maximum of 3% (2% due to imprecisions in sample volumes and 1% due to the apparatus). The total anisotropy percentage is given by P% = ((K1/K3)-1) x 100, or, to isolate the paramagnetic component, Pp% = ((K1-D)/ (K3-D)-1) x 100, which is the ratio corrected for the diamagnetic component D, considered as constant and isotropic (D = -1.4 x 10-5 S.I.). This assumes almost constant modal compositions and an extremely weak intrinsic anisotropy of quartz and feldspar (Bouchez et al., 187). The linear and planar anisotropy ratios are respectively denoted Lp% = ((K1-D)/(K2-D)-1) x 100 and Fp% = ((K2-D)/(K3-D)-1) x 100. RELATIONSHIP BETWEEN MAGNETIC FABRIC AND MINERAL FABRIC Various studies using independent measurement techniques (e.g. Heller, 173; Guillet et al., 183; Darrozes et al., 14) have demonstrated that in dominantly paramagnetic granites, the magnetic fabric is parallel to the mineral fabric. In biotite-bearing granites, the magnetic foliation and lineation correspond respectively to the orientation mean plane and zone axis of the biotite flakes. As the granites studied here have entirely magmatic microstructures, i.e. display no solid-state deformation features (Paterson et al., 18; Bouchez et al., 12), the foliation and lineation respectively correspond to the flattening plane and stretching direction acquired by the end of crystallization of the magma. The relationship between the magnetic anisotropy ratio and the finite strain suffered by the granite depends on the strain path and strain intensity ( see Arbaret et al., this volume) and several parameters intrinsic to the rock such as the shape of the iron-bearing minerals, their magneto-crystalline

3 HOMOGENEITY OF GRANITE FABRICS 115 anisotropy, the relative abundance of mineral species and the grain size homogeneity. A quantitative evaluation of the finite strain undergone by the crystallizing magma is therefore difficult, but a qualitative evaluation may be proposed: Pp < 3% corresponds to weakly oriented granites, 3%< Pp< 7% corresponds to a clearly visible planar orientation of the rock, and Pp >7% characterizes a strongly oriented granite. SAMPLING Three sites in fresh granites displaying different mean bulk magnetic anisotropies were sampled (Fig. 1). The least anisotropic site (Pp = 2.3%) is located in a quarry in the Sidobre (SI) monzogranite pluton (French Massif Central); the second site (Pp = 3.3%) is on a recently eroded glacial surface in the monzogranitic part of the Bassiès (BA) pluton (French Pyrenees); and the most anisotropic (Pp = 5.%) is on a recently eroded glacial surface in the Trois-Seigneurs (TS) granodiorite (French Pyrenees). These three Variscan plutons were recently studied for their magnetic structures (Moisy, 13; Darrozes et al., 14; Gleizes et al., 11; Leblanc et al., in press). BASSIES N P F SIDOBRE BA 4 km FRANCE Toulouse SI TROIS-SEIGNEURS N P Y Y SPAIN R E N E R E N E E E S S TS Figure 1. The Sidobre, Bassiès and Trois-Seigneurs plutons. The studied sites (SI, BA and TS) are located on the corresponding maps of magnetic lineations. NPF: North Pyrenean fault. All three plutons.are at the same scale. At each site, two square grids with 7 x 7 sampling points, plus one outside (Fig. 2), i.e. 50 sampling points, were core-drilled. There were a dekametric grid (dam) for which the cores are ten metres apart, and a metric grid (m) included in the large grid, for which the cores were taken one metre apart. Each core yielded two specimens, about 11 cm 3 in

4 11 Ph. OLIVIER ET AL. volume each, which were separately measured. The mean of both measurements characterizes each core. The mean grain areas, determined on thin sections (Fig. 3), are the following: SI: 0. mm 2 (standard deviation σ = 2. mm 2 ); BA: 0.5 mm 2 (σ = 1.2 mm 2 ); TS: 0.2 mm 2 (σ = 0.3 mm 2 ) SI TS BA m 50 Figure 2. Sampling grids. The small grid is located close to point 12 of the large grid for SI, point 32 for BA and point 17 for TS. DATA The raw data, calculated from the sets of 50 cores, are given for each grid in Tables 1 and 2 for the mean orientations of the lineations and foliations, Table 3 for the mean susceptibility magnitudes, and Table 4 for the mean anisotropy percentages. An elementary statistical analysis is given for each sample set, i.e. standard deviation (σ) and coefficient of variation (cv1 = σ/mean), providing an estimate of the within-grid variability. The within-core variability is estimated by the mean (mdsp) and standard deviation (σsp) of the departures between both specimens of the cores, and either, for the directional data, by the percentage of cores displaying a departure between both specimens from 0 to 20, or, for the scalar data, by a coefficient of variation cv2 = σsp/mean. The variability of the directional data is also represented by orientation in equal-area diagrams (Fig. 4). Spatial organization of the data is depicted on maps of figures 5 to 8 respectively corresponding to the lineations and foliations orientations, susceptibility and anisotropy magnitudes. On these maps, the fiftieth sample is omitted for the sake of simplification. Each data point represents the mean of the two specimens (22cm 3 ) of the core drilled at this point. LINEATIONS AND FOLIATIONS The main results arising from this study are i) the good clustering of the magnetic lineations, and to a lesser extent foliations, around a mean direction; ii) the similar orientations of these structures, for a given site, on both metre and dekametre scales.

5 HOMOGENEITY OF GRANITE FABRICS 117 Magnetic lineations K1 are the directional data which display the most homogeneous groupings (Table 1; Fig. 4). Their variability is weakest for TS and BA (Table 1, σ), Figure 3. Thin sections perpendicular to core axis (core diameter 25 mm), of representative textures of SI, BA and TS.

6 118 Ph. OLIVIER ET AL. slightly higher for SI, particularly on the large grid (σ = 25 ). The angular dispersion on the small grid is lower than on the large grid for all three sites. The angular departure between the mean lineations of both grids of a site is lowest for TS (2 : 20 /1 against 20 / 17 ), highest for BA ( : 47 / 41 against 5 / 35 ). TABLE 1. mean K1 and σ: mean and standard deviation of the lineations of 50 cores (100 specimens); mdsp, σsp: mean and standard deviation of the departures between the two specimens of a same core; 0-20 : percentage of cores for which departure between lineations of individual specimens lies in the range 0 to 20. mean K1 σ ( ) mdsp ( ) σsp ( ) 0-20 SI-dam 248 / SI-m 243 / BA-dam 47 / BA-m 5 / TS-dam 20 / TS-m 20 / The within-core variability of the lineations (Table 1) is low for TS (σsp = 7 - ; 88%-4% of the cores have an angular departure between individual k1 ranging from 0 to 20 ), and, to a lesser extent, for BA (σsp = 12 - ; 70%-0% of the cores have wellgrouped individual specimens), clearly higher for SI (σsp = ; 4%-50% of the cores have well-grouped individual specimens). In spite of these intrinsic variabilities, the overall lineation trends and plunges appear to be well organized, the transitions from one direction to another generally being gradual (Fig. 5). TABLE 2. mean K3 and σ: mean and standard deviation of the K3 of 50 cores (100 specimens); mdsp, σsp.: mean and standard deviation of the departures between the two specimens of a same core; 0-20 : percentage of cores for which departure between k3 of individual specimens lies in the range 0 to 20. mean K3 σ ( ) mdsp ( ) σsp ( ) 0-20 SI-dam 14 / SI-m 14 / BA-dam 17 / BA-m 173 / TS-dam 34 / TS-m 32 /

7 HOMOGENEITY OF GRANITE FABRICS 11 Large grid Small grid SI13 SI14 BA15 SI7 BA3 SI2 BA1 SI8 BA4 BA BA10 TROIS-SEIGNEURS BASSIES SIDOBRE TS5 TS11 TS12 TS TS17-18 TS17-18 Figure 4. Principal axes of the ellipsoid of magnetic susceptibility plotted on stereonets (equal area, lower hemisphere). Squares: K1, magnetic lineation; triangles: K2; circles: K3, normal to the magnetic foliation. Open symbols: values of the 50 cores (100 specimens) per grid; black symbols: grid mean. - For SI, the lineations fluctuate between NE-SW and ENE-WSW, with shallow plunges. This is particularly apparent on the small grid which displays a sigmoidal pattern, but harder to characterize precisely on the large grid; - For BA, the lineations plunge moderately to the NE, displaying gradual fluctuations which outline a sigmoidal pattern, between NNE and ENE in trends on the large grid, compared with NE to ENE on the small grid; - For TS, the lineations vary on both grids between NW and W in trends, with shallow plunges generally to the NW.

8 120 Ph. OLIVIER ET AL. K3 axes, the normal to magnetic foliations, are generally more variable than K1 axes, as attested by σ (Table 2) ranging from 15 (BA-m) to 52 (SI-m). For SI, and to a lesser extent for TS, this variability is partly due to the zonal distribution of the foliations around a mean direction close to the mean lineation (Fig. 4), a phenomenon frequently observed in granitoids. For SI, this results in a significant difference in plunge between the mean K3 axes of both grids (30 : 14 / 52 against 14 / 22 ), while, for TS both mean K3 axes are very close (34 / 41 against 32 / 42 ). For BA, which displays no zonal distribution of K3, mean K3 of both grids are also similar (17 / 35 against 173 / 30 ). Within-core variability (Table 2) is the lowest for BA (σsp = 15-7 ; 8%-88% of the cores have an angular departure between individual k3 from 0 to 20 ) and TS (σsp = ; 8%-88% of the cores have well-grouped individual specimens), more important for SI (σsp = ; 32%-28% of the cores have well-grouped individual specimens). The latter variability may account for the poor organization of the foliations in map view for SI, for which however, a NE-SW trend is visible, particularly on the small grid (Fig. ). For the other two sites, the clustering of the foliations around a mean value is obvious. For BA, the dominantly E-W striking, northward steeply dipping foliations vary rather regularly between ENE-WSW and ESE-WNW for the large grid, and between ENE-WSW and E-W for the small grid. For TS, the dominantly SE-NW striking, southwestward moderately dipping foliations fluctuate between E-W and SE-NW, with rather sharp transitions from one direction to the other on the large grid, changing more gradually between ESE-WNW and SE-NW on the small grid. TABLE 3. Km, MKm, mkm and σ: mean magnetic susceptibility, maximum, minimum and standard deviation of 50 cores (100 specimens); cv1 = (σ / Km) x 100; mdsp, σsp: mean and standard deviation of the departures between both specimens of a same core; cv2 = (σsp / Km) x 100. Km (10-5 SI) Mkm (10-5 SI) mkm (10-5 SI) σ cv1 mdsp σsp cv2 SI-dam SI-m BA-dam BA-m TS-dam TS-m MAGNETIC SUSCEPTIBILITY Magnetic susceptibility magnitudes are very homogeneous for the three sites, as shown by the low values of σ and the coefficient of variation cv1 (σ/km) (Table 3). On the large grid, the variability is about the same for SI as for TS (cv1 = 10% and 11% respectively), while it is much lower for BA (cv1 = 5%). On the small grids SI, BA and TS display similar variabilities (respectively %, % and 8%), lower than those of the large grid, except for BA.

9 HOMOGENEITY OF GRANITE FABRICS m 1 m Large grid Small grid BASSIES SIDOBRE TROIS-SEIGNEURS Figure 5. Magnetic lineations maps. The values at the tip of the arrows are the plunges. Black arrows correspond to the cores with an angular departure between individual k1 from 0 to 20 ; higher than 20 for the white arrows. The square in the large grid represents the position of the small grid. Within-core variability (Table 3) is rather high, with respect to the within-grid variability, for the three sites as shown by cv2 (= σsp/km) spanning from 4% for BA-dam to 7% for SI-dam and TS-m. Despite these variabilities, the susceptibility magnitudes generally constitute several rather extensive zones of similar values (Fig. 7), and isolated values very different from their neighbours are rare. The boundaries of the main isovalue zones display orientations that do not depend on the rock magnetic orientation, except for SI-dam.

10 122 Ph. OLIVIER ET AL. ANISOTROPY OF THE MAGNETIC SUSCEPTIBILITY The total magnetic anisotropy percentages are moderately homogeneous for the different grids of the three sites. This is reflected by the rather large variation of cv1 from 17% (BA-m) to 3% (SI-m) (Table 4). The mean total anisotropy percentages are equal on both the large and small grids for SI and BA, while TS displays a larger Pp% on the small grid than on the large grid. Amongst the large grids, SI displays the highest variability, while BA and TS have weaker and similar variabilities. For the metric grids, SI is again the most variable, with a rather higher value than on the large grid, followed by TS displaying about the same value as on the large grid, and BA displaying a value weaker than on the large grid. Table 4. Pp, MPp, mpp and σ: mean total anisotropy ratio, maximum, minimum and standard deviation of 50 cores (100 specimens); cv1 = (σ / Pp) x 100; mdsp, σsp: mean and standard deviation between both specimens of a same core; cv2 = (σsp / Pp) x 100. Fp, Lp: mean planar and linear anisotropy ratio of 50 cores (100 specimens). Pp MPp mpp σ cv1 mdsp σsp cv2 SI-dam SI-m BA-dam BA-m TS-dam TS-m Within-core variability is rather strong, compared to the within-grid variability, for SI (cv2 = 32%-0%), moderate for BA (cv2 = 14%-%) and TS (cv2 = 12%-8%) (Table 4). Despite these variabilities a zoned organization of Pp% appears on the different grids, with several large isovalue zones, and few isolated values (Fig. 8). The boundaries of the main zones of isovalues display slight tendencies to be oriented, for example parallel to N-S for TS-dam, and parallel to E-W on BA-m. Fp Lp DISCUSSION Homogeneity of the directional data, particularly the lineations, and their stability between scales, constitute undoubtedly the most spectacular result of this study. As an example, for the Sidobre site, the most variable amongst the three sites, the mean lineation is 248 / for the dekametric grid, and 243 / for the metric grid. Comparison between the lineations patterns at the grid scale (Fig. 5) and at the pluton scale (Fig. 1) also illustrates this homogeneity. Spatial variations of the directional data constitute an other important aspect of this study which is discussed hereafter with regard to intrinsic and external parameters. Scalar data, particularly the susceptibility magnitudes, also display bulk homogeneity. Spatial variations of these data, which are less clearly organized than for the directional data, are then briefly discussed.

11 HOMOGENEITY OF GRANITE FABRICS 123 DIRECTIONAL DATA In map view (Figs. 5 and ) the directional data generally vary gradually, but abrupt variations appear locally. These abrupt variations with respect to both the general trend and the neighbouring values, may be induced by different "intrinsic" causes, i.e. sample scale heterogeneities: - i) grain size heterogeneity may explain some large differences between directional data of the two specimens of a given core (Tables 1 and 2: σsp), particularly in SI, which has the largest and most heterogeneous grain-size (Fig. 3). In this porphyritic granite, K-feldspar megacrysts may perturb the biotite sub-fabric, and therefore the magnetic fabric around them; - ii) the deformation regime of the magma may cause specific variations. For example, the tendency for the foliations to be distributed in a zone around a mean direction as in SI and to a lesser extent TS (Fig. 4: partial girdles of K3), may correspond to a strong linear component of the fabric. This is observed at the grid scale, but is also evidenced at the core scale, particularly for SI where about 15% of the cores display very different plunges for both k3 of the same core. Note that the shape of the magnetic ellipsoid being dependent on the mineral marker, here the biotite flakes, the Lp%/Fp% ratio (see Table 4) cannot be used directly to determine whether the fabric of the rock is linear or planar; - iii) the anisotropy magnitude is certainly, in most cases, the controlling factor for orientation variations, i.e. the lower the anisotropy, the less well defined will be the directional data. Within a given sample, this is reflected by k1-k2 axes permutations for one or two specimen(s) of the same core when the linear contribution of the total anisotropy is especially weak: this occurs in 5% of the samples in SI, the least anisotropic site. Similarly k2-k3 axes permutations are observed for specimens having particularly weak planar anisotropy: this concerns 8% of the samples in SI, 2% in TS and 1% in BA. These different factors of within-core variability may combine, as for SI, the least anisotropic site, making variations at the grid scale difficult to interpret. On the contrary, when within-core variability is low, as for BA and TS, the variations of the directional data probably represent grid scale variations of the magmatic fabric. These variations are generally gradual, and some of them describe sigmoidal features, particularly in BA. For this site, the lineations mainly trend between N40 and N70 in the small grid, and between N20 and N70 in the large grid. This accounts for the rather high difference ( ) between the lineation means of both grids and indicates that the small grid represents a subdomain of the larger sigmoid evidenced in the large grid. In other words, the lineation strikes and plunges in the small grid do not reproduce the pattern observed in the large grid. It is proposed therefore that the wavelength of the undulations are larger than the size of the small grid. Conversely, for TS, the important angular departures sometimes observed between neighbouring directional data, particularly foliations in the large grid (Fig. ), could be due to undulations with wavelengths smaller than the mesh of the grid. It is tempting to ascribe the sigmoidal structures that appear on both the lineations and foliations maps to C/S patterns like those encountered in solid-state deformed rocks (Berthé et al., 17). For such an interpretation, a C orientation close to a shear plane must be identified along which the rock should be more strained than along a S orientation parallel to the finite stretch. But our maps do not display correlation between the highest magnitudes of Pp and particular structural directions, and so we cannot certainly identify C and S in our magnetic orientation patterns. Moreover, there is no evidence on

12 124 Ph. OLIVIER ET AL. the stereonets for bimodal distribution of the structural data. Alternatively, the waviness of the mineral fabric could be related to mechanical interactions in the crystal mush during crystallization of the magma, resulting in a fluctuation around a mean direction of stretch. Hence, even if the non-random nature of these sigmoids appears to be established, their significance is not clearly understood. 10 m 1 m Large grid Small grid BASSIES SIDOBRE TROIS-SEIGNEURS Figure. Magnetic foliations maps. The values are the foliation dips. Solid lines correspond to the cores having an angular departure between individual k3 ranging from 0 to 20 ; larger than 20 represented by open lines.

13 HOMOGENEITY OF GRANITE FABRICS 125 SCALAR DATA The magnetic susceptibility magnitudes are characterized by a rather strong clustering of the data around their means, while the anisotropy values are more variable (see cv1, Table 3 and 4). The scalar data have in common their rather heterogeneous spatial organizations. They are distributed, in map view, in different zones of isovalues, which might correspond to geologically organized domains having homogeneous biotite contents (K), and homogeneous strains (Pp). Both for K and Pp, and for the three sites, several values of the small grids fall outside the range of values of the corresponding zones in the large grids, thus indicating heterogeneities in these zones. The most striking example is given by TS for which the average Pp of the small grid (.1%) is clearly higher than for the large grid (5.1%). These heterogeneities may be attributed to the within-core variability, which is rather high compared with the grid scale variability, in particular for Pp for which cv2 ranges from 8% (TS-m) to 0% (SI-m). The within-core variability is probably mostly due to grain size and distribution heterogeneities, particularly for SI for which a single megacryst of K-feldspar can induce in a specimen important changes in biotite content and fabric. However, grain heterogeneity cannot account for all the variability of the scalar data. For instance, TS, the site having the most homogeneous grain size, displays a within-core variability similar to SI and BA for K values, and similar to BA for Pp values. It may also be noted that it is difficult to correlate the isovalue lines, either between K and Pp, or between one of these parameters and the structural trends, or sigmoids that are observed on the lineation and foliation maps. For example, although in the large grid of SI the K-value zones display a NE-SW trend, more-or-less parallel to the lineation trend (Fig. 5), this correlation is not found at the metre scale, where the isovalue lines of K display an approximately N-S trend. Nevertheless, the presence of rather large zones of isovalue tends to demonstrate the robustness of zonation in K and Pp, on both scales. The spatial organization of K is thus attributed to rock areas of several metres to dekametres, either enriched or depleted in biotite. This results either from an incomplete homogeneisation of the magma, or from a local partition between the solid fraction and the melt, by the end of crystallization. The spatial organization of Pp values probably indicates strain heterogeneities at the grid scales. As no clear correlation is evident between the orientation of the zones of anisotropy and the structural trends, no kinematic interpretation is proposed. CONCLUSION This study, which demonstrates the homogeneity of the directional data from a few metres to several tens of metres, complements fabric analyses previously performed in granites for which the directional data are often remarkably homogeneous in orientation on the scale of the whole pluton. This study also deals with the organization and random variations that appear in map views for both directional and scalar data. Transitions between neighbouring sampling points are generally gradual, and display consistent spatial organizations, such as sigmoids for directional data. Random variations are represented by few isolated and scattered values, at variance with the general trend. These are

14 12 Ph. OLIVIER ET AL. taken to be a function of heterogeneous grain size and local inhomogeneities of the magma. K 10 m 1 m Large grid Small grid BASSIES SIDOBRE K(10-5 SI) TROIS-SEIGNEURS Figure 7. Magnetic susceptibility magnitude maps. Vertical bars on both sides of the scale include 0% of the values, and the position of the mean of the corresponding grid. More extensive statistical analyses, based on these raw data, are now necessary in order to quantify better the spatially organized variations, like the geometry of sigmoidal features (K1, K3), the localisation of the strain (P%), or petrographic zoning (K).

15 HOMOGENEITY OF GRANITE FABRICS 127 Pp% 10 m Large grid 1 m Small grid Pp% BASSIES SIDOBRE TROIS-SEIGNEURS Figure 8. Anisotropy of magnetic susceptibility maps. Vertical bars on both sides of the scale include 0% of the values, and the position of the mean of the corresponding grid. ACKNOWLEDGEMENTS We are greatly indebted to Ed. Stephens and C. Teyssier for reviewing and improving this paper.